Aim: To determine whether levels of daytime physical activity are associated with sleep duration and night waking in children assessed using accelerometry, and if these associations change over time. Methods: 24-hour accelerometry data were obtained from 234 children at 3, 5 and 7 years of age for at least 5 days at each time. Sleep duration was estimated using the Sadeh algorithm. Time spent in sedentary, light and moderate-vigorous (MVPA) activity was established using published cut-points. Appropriate statistical techniques were utilised to account for the closed nature of the data (24-hour periods). Results: Time spent asleep was related more to sedentary or light activity and not to MVPA. The most active (95th percentile) children spent 55–84 fewer minutes asleep and 16–19 more minutes awake at night compared to the least active (5th percentile) children. Children with later bedtimes slept less at night (30–40 minutes) and undertook more sedentary (10–15 minutes) but also more light (18–23 minutes) activity during the day. However, no differences in MVPA were apparent according to bedtime. Children slept slightly less on weekend nights (11 minutes) compared with week-nights, but only at 3 years of age. Most relationships were broadly similar at 3, 5 and 7 years of age. Conclusion: Children who are more physically active during the day have shorter total sleep time and are more awake at night than less active children. The protective effect of sleep on obesity does not appear to be mediated by increased physical activity.